The Research Of Detecting The Number Of People In Lift Car Based On Adaboost Algorithm And Background Subtraction Method

With the continuously economic improvement of the modern society, more and more high-rise buildings come out from the ground, which makes a rapid progress of the elevator industry. So the safety problems arouse more people’s concern. The overload limit device is designed generally by the gravity sensor. Once the gravity sensor goes wrong, it may causes serious safety accidents. So, based on improving the performance of the gravity sensor, it can strengthen the overload limit device’s safety performance based on controlling the number of people inside the lift car.This paprer focuses on how to rapidly and accurately detect the human face from the image monitored by the video on the upper computer, and then get the number of the people inside the elevator. In a word, the study of number of people detection indirectly turns into the study of face detection.Firstly, this paper introduces the background and the meaning of the research and the status of face detection research now. Combine the main methods of the face detection’s characteristics and application requirements, it selects Adaboost algorithm as the main research method. Then it introduces the process of face detection mechanism and some related basic concepts, including the gray image, the integral image, Haar features, etc, which laid sound foundation for the experiment.This article designed an face detection experiment based on Adaboost algorithm, according to the current research on positive face classifier. It tested the images monitored on the video of the elevator. Experiments show that the positive face classifier is not suitable for special environment. Aimming at the specific shooting angle, this article has collected and made special angel face samples for training the classifier, and experiment on this classifier.Aimming at the fixed application background, the background subtraction method was applied to the process of face detection. According to the background subtraction method, it can exclude the disturbance and decrease the false-alarm rate.Finally, this paper tested on face detection algorithm and designed a simple and feasible face detection software on the basis of above algorithms by MFC. The experimental result is good. This face detection algorithm has a feasible prospect of application.